Entropy, heat and microorganism transport analysis of shape-based hybrid nanofluid squeezing flow: An ANN-based numerical approach

This study numerically investigates entropy generation and uses an artificial neural network (ANN) to predict heat and microorganism transport in hybrid nanofluids (HNFs) containing multiple nanometals with distinct geometries: TiO 2 (blade), Cu (brick), Fe 3 O 4 (cylinder), MoS 2 (platelet), and Ag (spherical), individually combined with spherical Al 2 O 3 nanoparticles in a blood-based medium containing microorganisms flowing between two parallel plates under couple stress and magnetohydrodynamic effects. A novel thermophysical property model is proposed for these HNFs. The governing equations are transformed into dimensionless form and solved using a shooting method coupled with a sixth-order Runge–Kutta scheme. The effects of physical parameters and nanoparticle shapes on velocity, temperature, microorganism concentration, and entropy generation are presented graphically. Results show that TiO 2 + Al 2 O 3 /blood exhibits the highest thermal performance, whereas MoS 2 + Al 2 O 3 /blood shows the lowest. The ANN accurately predicts temperature and microorganism distributions, demonstrating excellent agreement with numerical results. Its performance is evaluated using error histograms, regression analysis, mean squared error, and function-fit analysis. The ANN achieves a regression coefficient of R = 1 and a mean squared error of 4.2863 × 10 −11 , indicating excellent prediction accuracy and convergence. The couple stress parameter ( β ) has the strongest influence on entropy generation, increasing it by 21440.50%, while the Hartmann number (Ha) produces the most significant effect on microorganism concentration, reducing it by 0.448%. These findings have potential applications in biomedical systems, microfluidics, electronic cooling, and enhanced heat-transfer devices.